Term Rewriting
Rule-based rewriting
Rewrite rules match and transform an expression. A rule is written using either the @rule
macro or the @acrule
macro. It creates a callable Rule
object.
Basics of rule-based term rewriting in SymbolicUtils
Here is a simple rewrite rule, that uses formula for the double angle of the sine function:
using SymbolicUtils
@syms w z α::Real β::Real
(w, z, α, β) # hide
r1 = @rule sin(2(~x)) => 2sin(~x)*cos(~x)
r1(sin(2z))
# output
2sin(z)*cos(z)
The @rule
macro takes a pair of patterns – the matcher and the consequent (@rule matcher => consequent
). If an expression matches the matcher pattern, it is rewritten to the consequent pattern. @rule
returns a callable object that applies the rule to an expression.
~x
in the example is what is a slot variable named x
. In a matcher pattern, slot variables are placeholders that match exactly one expression. When used on the consequent side, they stand in for the matched expression. If a slot variable appears twice in a matcher pattern, all corresponding matches must be equal (as tested by Base.isequal
function). Hence this rule says: if you see something added to itself, make it twice of that thing, and works as such.
If you try to apply this rule to an expression with triple angle, it will return nothing
– this is the way a rule signifies failure to match.
r1(sin(3z)) === nothing
# output
true
Slot variable (matcher) is not necessary a single variable
r1(sin(2*(w-z)))
# output
2cos(w - z)*sin(w - z)
but it must be a single expression
r1(sin(2*(w+z)*(α+β))) === nothing
# output
true
Rules are of course not limited to single slot variable
r2 = @rule sin(~x + ~y) => sin(~x)*cos(~y) + cos(~x)*sin(~y);
r2(sin(α+β))
# output
sin(β)*cos(α) + cos(β)*sin(α)
If you want to match a variable number of subexpressions at once, you will need a segment variable. ~~xs
in the following example is a segment variable:
@syms x y z
@rule(+(~~xs) => ~~xs)(x + y + z)
# output
3-element view(::Vector{Any}, 1:3) with eltype Any:
z
y
x
~~xs
is a vector of subexpressions matched. You can use it to construct something more useful:
r3 = @rule ~x * +(~~ys) => sum(map(y-> ~x * y, ~~ys));
r3(2 * (w+w+α+β))
# output
4w + 2α + 2β
Notice that the expression was autosimplified before application of the rule.
2 * (w+w+α+β)
# output
2(2w + α + β)
Predicates for matching
Matcher pattern may contain slot variables with attached predicates, written as ~x::f
where f
is a function that takes a matched expression and returns a boolean value. Such a slot will be considered a match only if f
returns true.
Similarly ~~x::g
is a way of attaching a predicate g
to a segment variable. In the case of segment variables g
gets a vector of 0 or more expressions and must return a boolean value. If the same slot or segment variable appears twice in the matcher pattern, then at most one of the occurrence should have a predicate.
For example,
using SymbolicUtils
@syms a b c d
r = @rule ~x + ~~y::(ys->iseven(length(ys))) => "odd terms";
@show r(a + b + c + d)
@show r(b + c + d)
@show r(b + c + b)
@show r(a + b)
# output
r(a + b + c + d) = nothing
r(b + c + d) = "odd terms"
r(b + c + b) = nothing
r(a + b) = nothing
Associative-Commutative Rules
Given an expression f(x, f(y, z, u), v, w)
, a f
is said to be associative if the expression is equivalent to f(x, y, z, u, v, w)
and commutative if the order of arguments does not matter. SymbolicUtils has a special @acrule
macro meant for rules on functions which are associate and commutative such as addition and multiplication of real and complex numbers.
using SymbolicUtils
@syms x y z
acr = @acrule((~a)^(~x) * (~a)^(~y) => (~a)^(~x + ~y))
acr(x^y * x^z)
# output
x^(y + z)
although in case of Number
it also works the same way with regular @rule
since autosimplification orders and applies associativity and commutativity to the expression.
Example of applying the rules to simplify expression
Consider expression (cos(x) + sin(x))^2
that we would like simplify by applying some trigonometric rules. First, we need rule to expand square of cos(x) + sin(x)
. First we try the simplest rule to expand square of the sum and try it on simple expression
using SymbolicUtils
@syms x::Real y::Real
sqexpand = @rule (~x + ~y)^2 => (~x)^2 + (~y)^2 + 2 * ~x * ~y
sqexpand((cos(x) + sin(x))^2)
# output
sin(x)^2 + 2sin(x)*cos(x) + cos(x)^2
It works. This can be further simplified using Pythagorean identity and check it
pyid = @rule sin(~x)^2 + cos(~x)^2 => 1
pyid(cos(x)^2 + sin(x)^2) === nothing
# output
true
Why does it return nothing
? If we look at the rule, we see that the order of sin(x)
and cos(x)
is different. Therefore, in order to work, the rule needs to be associative-commutative.
acpyid = @acrule sin(~x)^2 + cos(~x)^2 => 1
acpyid(cos(x)^2 + sin(x)^2 + 2cos(x)*sin(x))
# output
1 + 2sin(x)*cos(x)
It has been some work. Fortunately rules may be chained together into more sophisticated rewriters to avoid manual application of the rules.
Composing rewriters
A rewriter is any callable object which takes an expression and returns an expression or nothing
. If nothing
is returned that means there was no changes applicable to the input expression. The Rules we created above are rewriters.
The SymbolicUtils.Rewriters
module contains some types which create and transform rewriters.
Empty()
is a rewriter which always returnsnothing
Chain(itr)
chain an iterator of rewriters into a single rewriter which applies each chained rewriter in the given order. If a rewriter returnsnothing
this is treated as a no-change.RestartedChain(itr)
likeChain(itr)
but restarts from the first rewriter once on the first successful application of one of the chained rewriters.IfElse(cond, rw1, rw2)
runs thecond
function on the input, appliesrw1
if cond returns true,rw2
if it returns falseIf(cond, rw)
is the same asIfElse(cond, rw, Empty())
Prewalk(rw; threaded=false, thread_cutoff=100)
returns a rewriter which does a pre-order (from top to bottom and from left to right) traversal of a given expression and applies the rewriterrw
.threaded=true
will use multi threading for traversal. Note that ifrw
returnsnothing
when a match is not found, thenPrewalk(rw)
will also return nothing unless a match is found at every level of the walk. If you are applying multiple rules, thenChain
already has the appropriate passthrough behavior. If you only want to apply one rule, then consider usingPassThrough
.thread_cutoff
is the minimum number of nodes in a subtree which should be walked in a threaded spawn.Postwalk(rw; threaded=false, thread_cutoff=100)
similarly does post-order (from left to right and from bottom to top) traversal.Fixpoint(rw)
returns a rewriter which appliesrw
repeatedly until there are no changes to be made.PassThrough(rw)
returns a rewriter which ifrw(x)
returnsnothing
will instead returnx
otherwise will returnrw(x)
.
Chaining rewriters
Several rules may be chained to give chain of rules. Chain is an array of rules which are subsequently applied to the expression.
To check that, we will combine rules from previous example into a chain
using SymbolicUtils
using SymbolicUtils.Rewriters
@syms x
sqexpand = @rule (~x + ~y)^2 => (~x)^2 + (~y)^2 + 2 * ~x * ~y
acpyid = @acrule sin(~x)^2 + cos(~x)^2 => 1
csa = Chain([sqexpand, acpyid])
csa((cos(x) + sin(x))^2)
# output
1 + 2sin(x)*cos(x)
Important feature of Chain
is that it returns the expression instead of nothing
if it doesn't change the expression
Chain([@acrule sin(~x)^2 + cos(~x)^2 => 1])((cos(x) + sin(x))^2)
# output
(sin(x) + cos(x))^2
it's important to notice, that chain is ordered, so if rules are in different order it wouldn't work the same as in earlier example
cas = Chain([acpyid, sqexpand])
cas((cos(x) + sin(x))^2)
# output
sin(x)^2 + 2sin(x)*cos(x) + cos(x)^2
since Pythagorean identity is applied before square expansion, so it is unable to match squares of sine and cosine.
One way to circumvent the problem of order of applying rules in chain is to use RestartedChain
using SymbolicUtils.Rewriters: RestartedChain
rcas = RestartedChain([acpyid, sqexpand])
rcas((cos(x) + sin(x))^2)
# output
1 + 2sin(x)*cos(x)
It restarts the chain after each successful application of a rule, so after sqexpand
is hit it (re)starts again and successfully applies acpyid
to resulting expression.
You can also use Fixpoint
to apply the rules until there are no changes.
Fixpoint(cas)((cos(x) + sin(x))^2)
# output
1 + 2sin(x)*cos(x)